Machine learning and types

"Machine learning is field of study that gives computers the ability to learn about being explicitly programmed.", Arthur Samuel(1959)

Machine Learning
- Supervised learning(x input --> y labeld output)
    - Regression : predicting many possible numbs
    - Classification : predicting a few numbers like categories
- Un-supervised learning(x input but no right answers y)
    - Clustering
    - Anomaly detection
    - Dimensionality reduction
- Recommend system
- Reinforcement learning

 

What is machine learning

Machine learning is the field of solving the problem with computer and data. It is a part of AI.

 

Application variations

The function of ML depends on how programmers apply. If input is email and output tells whether it is spam or not, then this is spam filter. If its input is English and output is Spanish, then this is machine translation.

 

Types of ML

There are some different types of machine learning, for example, supervised learning, unsupervised learning, recommend system , reinforcement learning and etc.

 

What is Supervised Learning

Supervised learning is the method computer learns from given correct answers. Programmer give two datas, input x and output y(right answers). Y is labeled datas that represents whether it is correct or not.

 

Types of supervised learning

The major types of supervised learning algorithms are 1)Regression and 2)Classification.

Regression helps predicting many possible outputs. It is good for predicting number. The example can be price prediction.

On the other hand, Classification is for predicting categories. It can handle small number of possible outputs. For example, answering the image is whether about cat or dog, detecting breast cancer from x-rays and so on.

Regression is good at handling multiple outputs.

Classification works well with categorizing

 

What is Unsupervised learning

Unsupervised learning is the way computer learn without any given answers(No supervising). Programmer provides only input x but not labeld output y. Instead, computer itself find interesting pattern or structure from unlabeled datas.

 

Types of unsupervised learning algorithms

There are many types of unsupervised learning such as Clustering, Anomaly detection and Dimensionality reduction.

Clutering helps grouping similar datas together. Google news, DNA datas are real-life examples. Anomaly detection finds data points which has unusual pattern. Dimensionality reduction is for compressing data using fewer numbers.